Joint training with local soft attention and dual cross-neighbor label smoothing for unsupervised person re-identification  

作　　者：Qing Han Longfei Li Weidong Min Qi Wang Qingpeng Zeng Shimiao Cui Jiongjin Chen 

机构地区：[1]School of Mathematics and Computer Sciences,Nanchang University,Nanchang 330031,China [2]Institute of Metaverse,Nanchang University,Nanchang 330031,China [3]Jiangxi Key Laboratory of Smart City,Nanchang 330031,China

出　　处：《Computational Visual Media》2024年第3期543-558,共16页计算可视媒体（英文版）

基　　金：supported by the National Natural Science Foundation of China under Grant Nos.62076117 and 62166026;the Jiangxi Key Laboratory of Smart City under Grant No.20192BCD40002;Jiangxi Provincial Natural Science Foundation under Grant No.20224BAB212011.

摘　　要：Existing unsupervised person re-identification approaches fail to fully capture thefine-grained features of local regions,which can result in people with similar appearances and different identities being assigned the same label after clustering.The identity-independent information contained in different local regions leads to different levels of local noise.To address these challenges,joint training with local soft attention and dual cross-neighbor label smoothing(DCLS)is proposed in this study.First,the joint training is divided into global and local parts,whereby a soft attention mechanism is proposed for the local branch to accurately capture the subtle differences in local regions,which improves the ability of the re-identification model in identifying a person’s local significant features.Second,DCLS is designed to progressively mitigate label noise in different local regions.The DCLS uses global and local similarity metrics to semantically align the global and local regions of the person and further determines the proximity association between local regions through the cross information of neighboring regions,thereby achieving label smoothing of the global and local regions throughout the training process.In extensive experiments,the proposed method outperformed existing methods under unsupervised settings on several standard person re-identification datasets.

关 键 词：person re-identification(Re-ID) unsupervised learning(USL) local soft attention joint training dual cross-neighbor label smoothing(DCLS) 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]